New variation on the equation system solver.

Much simpler to understand. Basically have a variable assignment which
is dynamic and updates with the strategy changes. Similarly have
strategy changes easily invalidate the variable assignment.

This makes them strongly inter-dependent, but simplifies the
implementation considerably. Proving it should be easier like this, too.

1 files changed, 67 insertions, 0 deletions

diff --git a/TODO.org b/TODO.org
new file mode 100644
index 0000000..bb4d349
--- /dev/null
+++ b/TODO.org
@@ -0,0 +1,67 @@
+* TODO Refactoring [1/4]
+- [X] Fix up logging
+- [ ] Separate classes into files
+- [ ] Put outer-loop in a class (out of main)
+- [ ] Give EquationSystem methods for creating assignments/strategies
+* TODO Verbosity [2/3]
+- [X] Wrap output streams [5/5]
+  - [X] strategies
+  - [X] fixpoint
+  - [X] debug
+  - [X] trace
+  - [X] info
+- [X] Command-line arguments [5/5]
+  - [X] strategies
+  - [X] fixpoint
+  - [X] debug
+  - [X] trace
+  - [X] info
+- [-] Provide output [0/5]
+  - [-] strategies [1/2]
+    - [ ] print the system and assignment, then the next strategy
+    - [X] print each strategy like this:
+      max(0, x, sub(y,2)) --[2]-> sub(y,2)
+  - [ ] fixpoint
+    print the system we're working with, then print the resultant assignment
+  - [ ] debug
+    print the initial system we're working with
+    print more general info in the main loop (number of loops, each assignment/strategy)
+  - [ ] trace
+    maybe we don't need this
+    would be something like "entering/exiting" everything mildly important
+  - [ ] info
+    print the initial system we're working with    
+* TODO make test [2/3]
+- [-] write test cases [1/4]
+  - [X] examples from paper
+  - [ ] best cases
+  - [ ] worst cases
+  - [ ] average cases
+- [X] write a script to run over each test case and return pass/fail
+- [X] consolidate into makefile as `make test`
+* TODO Parametrise EquationSystem for Variable, too [0/1]
+- [ ] Give EquationSystem a template parameter for Variable
+  I don't think this is necessary, but Thomas wants it.
+  For now I'm going to hold off on it, though.
+* TODO Min-cost flow problem [1/3]
+- [X] Research min-cost flow libraries
+  Potentially just use LEMON? It looks pretty decent.
+  LEMON it is!
+- [ ] Implement the min-cost flow operator MCF
+  It takes four construction-time arguments and one run-time argument.
+  Construction time:
+  - T (the template constraint matrix)
+  - A (the multiplicative modification factor)
+  - b (the additive modification factor)
+  - c (the bound in the guard)
+  Run-time:
+  - d (the abstract value to be operated on)
+- [ ] Update test cases to include MCF
+* TODO Presentation [0/2]
+- [ ] Simplify "Even Faster" solver for explanation
+- [ ] Make slides [0/5]
+  - [ ] Even Faster Solver
+  - [ ] Strategy improvement
+  - [ ] Example of scaling
+  - [ ] Explain solution
+  - [ ] Demonstrate implementation